| --- Log opened Mon May 06 00:00:46 2013 | ||
| mor1kx | [mor1kx] skristiansson pushed 8 new commits to master: https://github.com/openrisc/mor1kx/compare/40fad7fa4c42...591f41b398c2 | 19:17 |
|---|---|---|
| mor1kx | mor1kx/master 9fb753a Stefan Kristiansson: cappuccino/decode_execute: move all op control signals to one always block... | 19:17 |
| mor1kx | mor1kx/master 6bc6aeb Stefan Kristiansson: cappuccino/decode_execute: remove reset logic from execute_rfd_adr... | 19:17 |
| mor1kx | mor1kx/master 977646b Stefan Kristiansson: cappuccino/decode_execute: clear execute_rf_wb in pipeline_flush_i | 19:17 |
| stekern | bah, *on* pipeline_flush_i... | 19:19 |
| stekern | juliusb: just a heads up, I've been touching the ticktimer: https://github.com/openrisc/mor1kx/commit/591f41b398c2253da17b6634fa0805e7890bd1da | 19:19 |
| stekern | nothing really controversial, just a small optimisation, without making it to obfuscated (I hope) | 19:21 |
| stekern | s/to/too | 19:21 |
| @juliusb | stekern: nice work on the trimming :) | 21:27 |
| @juliusb | no problem with tick timer, I think that looks neater | 21:28 |
| @juliusb | I'm still struggling with my pronto fetch cache :( | 21:28 |
| @juliusb | i fix one corner case and it pops another up | 21:29 |
| @juliusb | the idea was to make it zero branch overhead but to do that means you get some bad paths through everything (branch address calculated via ALU, then fed back to fetch unit and then that is compared with the cached tag | 21:51 |
| @juliusb | ) | 21:51 |
| @juliusb | So I haven't been able to achieve that, but I have certainly given the option to spend a few flops and get a nice little speedup | 21:53 |
| --- Log closed Tue May 07 00:00:47 2013 | ||
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!